ClearML hits 1.0 and comes of age
May 3rd 2021 – With over 11 man-years of working, and tinkering, long into the night, I am pleased to announce we have hit version 1.0. Following quickly after the release of ClearML 0.17.5, we added the last remaining features we felt 1.0 needed. Namely multi-model support, as well as improved batch operations. With these in place, the choice was clear. The next version released should be the baseline moving forward.
Who is Allegro AI and ClearML’s vision
Allegro AI is the leading MLOps provider. Our ClearML platform makes ML and DL researchers more effective by giving them tools to manage their experiments, data, pipelines and model deployment. ClearML vision is to be the missing devops piece in your machine learning software stack. ClearML gives data scientists tools to manage experiments, orchestrate workloads, manage data operations and deploy models, all in a simple open-source tool. Integration is largely agnostic, and should work with whatever toolchain a team is using already.
Key Features of the 1.0 Release
Since release 0.16 there have been some notable main features which have landed. I don’t want to sit and enumerate all of them, but from talking with the team, here are the main ones that we are proud of;
- AWS EC2 Auto-Scaler service wizard and Service.
- Vastly reworked documentation, easing onboarding
- Batch operations, the ability to work with projects/tasks in a much more time efficient way
- Offline support for Tasks
- Kubernetes integration
- Docker support
Overall, I think the above features make for a very healthy 1.0 release. Of course, the final point to note with a 1.0 release is that starting from this point forwards, ClearML will strive not to break backwards compatibility.
Where can I get ClearML 1.0 ?
ClearML 1.0 is available for free now on github for self installation. For those who prefer a more managed solution, we also offer a hosted solution or enterprise. Further details can be found on our pricing page
Celebrating the Good
It has been a long and sometimes bumpy path to get us to version 1.0. There have been good times and some mis-steps. Every journey is different, and we have been helped along the way by having one of the nicest slack communities I have ever been a part of. This feels truly like a milestone. A birthday. I hope you will help us in celebrating it.